tokenizer_config.json

{"keep_accents": true, "do_lower_case": false, "model_max_length": 512, "tokenizer_class": "AlbertTokenizer"}